What the research says about who really buys in B2B
The B2B content brief usually names one reader. The research has said for fifty years that the decision belongs to a group, and that the group's size moves with the purchase rather than with the company.
That gap explains a familiar failure. A campaign performs well with its target persona and the deal still stalls, because the people who stalled it were never addressed.
The full citations are in the B2B Buying Behavior & the Buying Center research index. A sourcing note: the most-quoted modern B2B buying statistics come from organizations that sell B2B advertising or gate the underlying study. The papers below are the peer-reviewed base those figures sit on top of, and they are checkable.
The buying center is the unit of decision
Webster and Wind (1972) modeled industrial purchasing in the Journal of Marketing as an organizational decision process rather than an individual choice. Their contribution was to name the buying center: the set of people who participate in a purchase, occupying different roles with different stakes.
The model sorts the influences on that decision into environmental, organizational, interpersonal and individual layers. A content program that speaks only to the individual layer is addressing the smallest of the four.
This is the reason persona-driven B2B content underperforms its own engagement numbers. The persona reads it, likes it, and then has to carry the argument into a room containing people who did not.
Group size tracks the purchase, not the company
Lewin and Donthu (2005) meta-analyzed the organizational buying literature in the Journal of Business Research and found a strong relationship between purchase situation and buying center size, reporting a correlation of r = .47.
That number is worth sitting with, because it contradicts how most teams plan. Buying group size is not a fixed attribute of a customer segment or a deal tier. It expands with novelty and risk in the purchase itself.
The same product sold as a routine renewal and as a first-time category purchase involves different numbers of people. One needs a reminder. The other needs a case that survives being retold by someone who is not you.
| Purchase situation | What the research implies | Content consequence |
|---|---|---|
| Routine repurchase | Fewer participants, lower involvement | Reinforcement; low volume is fine |
| Modified rebuy | Moderate group, some new criteria | Comparison material, migration detail |
| New task purchase | Larger group, higher involvement | Depth across roles; the case must travel |
This also reframes a question teams ask about AI content: how much do we need? The answer scales with how unfamiliar the purchase is to the buyer, not with how much the deal is worth.
Disagreement inside the group is a stage, not an exception
Sheth (1973) built the point into his model of industrial buyer behavior. Participants arrive with different expectations of the same supplier, shaped partly by which information sources each one consulted, and conflict resolution is an explicit stage of the process.
Most B2B content is written as though the reader's agreement ends the matter. It does not. The champion has to win an internal argument you will never see, against objections raised by a finance lead or a security reviewer who read nothing you published.
Content that helps someone else make your case is a different artifact from content that makes the case to a reader. It is more specific, more willing to state limits, and more useful in a document someone else is assembling.
What changes when AI produces the content
Johnston and Lewin (1996) reviewed 165 articles across six marketing journals to integrate twenty-five years of research after the foundational models. The value of a review like that is separating findings that replicated from claims that were merely proposed, which is a useful discipline to borrow when evaluating current B2B statistics.
Applied to AI production, the research points at coverage and consistency rather than volume. A buying group encounters different pieces at different moments across a decision that can run months. Every piece is evidence about the same supplier.
Drift is the specific risk. When each session starts cold, the security page and the pricing explainer and the case study can each be individually good while disagreeing about what the product is. Copper Sun keeps brand facts and locked decisions available across projects so the tenth asset agrees with the first.
The test is simple enough to apply to a content plan: if the finance lead, the security reviewer and the end user each read only one of these pieces, would they come away with the same understanding? If not, more content will produce more contradictions.
Frequently Asked Questions
How many people are in a typical B2B buying group?
The research does not support one number, and that is the more useful answer. Lewin and Donthu (2005) found a strong relationship between purchase situation and buying center size, at r = .47. Novel, higher-risk purchases involve more participants; routine repurchases involve fewer. Any single figure describes a particular sample of purchase situations rather than a property of B2B buying.
Why cite papers from the 1970s instead of current B2B buying statistics?
The widely quoted modern figures are published by organizations that sell B2B advertising or keep the underlying research gated, which makes them hard to check. The academic work establishes the same mechanism through peer-reviewed studies anyone can verify. Where a vendor statistic agrees with this base, the base is the stronger citation.
Is this research still valid given how much B2B buying has changed?
The foundational models describe decision structure rather than any channel, which is why Johnston and Lewin could integrate twenty-five years of later work into the same framework, and why Lewin and Donthu could meta-analyze it in 2005. What changed is how buyers gather information. What did not change is that several people with different expectations still have to reach one decision.
What should a B2B content plan do differently?
Map the roles a decision actually passes through, then check coverage against that map rather than against a persona. Give the internal champion material that works when they are not in the room. The related argument about consistency across volume is in what the research says about how brands actually grow.